Somatostatin analogs
Somatostatin is a cyclic tetradecapeptide found both in the central nervous system and in peripheral tissues. It was originally isolated from mammalian hypothalamus and identified as an important inhibitor of growth hormone secretion from the anterior pituitary. Its multiple biological activities include inhibition of the secretion of glucagon and insulin from the pancreas, regulation of most gut hormones and regulation of the release of other neurotransmitters involved in motor activity and cognitive processes throughout the central nervous system (for review see Lamberts, Endocrine Rev., 9:427, 1988). Additionally, somatostatin and its analogs are potentially useful antiproliferative agents for the treatment of various types of tumors.
Natural somatostatin (also known as Somatotropin Release Inhibiting Factor, SRIF) of the following structure: EQU H-Ala.sup.1 -Gly.sup.2 -Cys.sup.3 -Lys.sup.4 -Asn.sup.5 -Phe.sup.6 -Phe.sup.7 -Trp.sup.8 -Lys.sup.9 -Thr.sup.10 -Phe.sup.11 -Thr.sup.12 -Ser.sup.13 -Cys.sup.14 -OH
was first isolated by Guillemin and colleagues (Bruzeau et al. Science, 179:78, 1973). It exerts its effect by interacting with a family of receptors. Recently five receptor subtypes, termed SSTR1-5, have been identified and cloned. In its natural form, somatostatin has limited use as a therapeutic agent since it exhibits two undesirable properties: poor bioavailability and short duration of action. For this reason, great efforts have been made during the last two decades to find somatostatin analogs that will have superiority in either potency, biostability, duration of action or selectivity with regard to inhibition of the release of growth hormone, insulin or glucagon.
Structure-activity relation studies, spectroscopic techniques such as circular dichroism and nuclear magnetic resonance, and molecular modeling approaches reveal the following: the conformation of the cyclic part of natural somatostatin is most likely to be an antiparallel .beta.-sheet; Phe.sup.6 and Phe.sup.11 play an important role in stabilizing the pharmacophore conformation through hydrophobic interactions between the two aromatic rings; the four amino acids Phe.sup.7 -Trp.sup.9 -Lys.sup.9 -Thr.sup.10 which are spread around the .beta.-turn in the antiparallel .beta.-sheet are essential for the pharmacophore; and (D)Trp.sup.8 is preferable to (L)Trp.sup.8 for the interactions with somatostatin receptor subtypes 2 through 5.
Nevertheless, a hexapeptide somatostatin analog containing these four amino acids anchored by a disulfide bridge: ##STR1## is almost inactive both in vitro and in vivo, although it has the advantage of the covalent disulfide bridge which replaces the Phe.sup.6 -Phe.sup.11 hydrophobic interactions in natural somatostatin.
Four main approaches have been attempted in order to increase the activity of this hexapeptide somatostatin analog. (1) Replacing the disulfide bridge by a cyclization which encourages a cis-amide bond, or by performing a second cyclization to the molecule yielding a bicyclic analog. In both cases the resultant analog has a reduced number of conformational degrees of freedom. (2) Replacing the original residues in the sequence Phe.sup.7 -(D)Trp.sup.8 -Lys.sup.9 -Thr.sup.10 with other natural or non-natural amino acids, such as replacing Phe.sup.7 with Tyr.sup.7 and Thr.sup.10 with Val.sup.10. (3) Incorporating additional functional groups from natural somatostatin with the intention that these new elements will contribute to the interaction with the receptor. (4) Eliminating one of the four amino acids Phe.sup.7 -(D)Trp.sup.8 -Lys.sup.9 -Thr.sup.10 with the assumption that such analogs would be more selective.
The somatostatin analog, MK-678: EQU cyclo (N-Me-Ala.sup.6 -Tyr.sup.7 -(D)Trp.sup.8 -Lys.sup.9 -Val.sup.10 -Phe)
is an example of a highly potent somatostatin analog designed using the first three approaches above (Veber, et al., Life Science, 34:371, 1984). In this hexapeptide analog, a cis-amide bond is located between N-Me-Ala and Phe.sup.11, Tyr.sup.7 and Val.sup.10 replace Phe.sup.7 and Thr.sup.10 respectively, and Phe.sup.11 is incorporated from natural somatostatin.
Another group of somatostatin analogs (U.S. Pat. Nos. 4,310,518 and 4,235,886) includes Octreotide: ##STR2## the only approved somatostatin analog currently available. It was developed using the third approach described above. Here, (D) Phe.sup.5 and the reduced C-terminal Thr.sup.12 -CH.sub.2 OH are assumed to occupy some of the conformational space available to the natural Phe.sup.6 and Thr.sup.12, respectively.
The compound TT-232: ##STR3## is closely related to Octreotide and is an example of implementing the fourth approach described above. The lack of Thr.sup.10 is probably responsible for its high functional selectivity in terms of antitumor activity.
These examples of highly potent somatostatin analogs suggest that the phenylalanines in positions 6 and 11 not only play an important role in stabilizing the pharmacophore conformation but also have a functional role in the interaction with the receptor. It is still an open question whether one phenylalanine (either Phe.sup.6 or Phe.sup.11) is sufficient for the interaction with the receptor or whether both are needed.
It is now known that the somatostatin receptors constitute a family of five different receptor subtypes (Bell and Reisine, Trends Neurosci., 16, 34-38, 1993), which may be distinguished on the basis of their tissue specificity and/or biological activity. Somatostatin analogs known in the art may not provide sufficient selectivity or receptor subtype selectivity, particularly as anti-neoplastic agents (Reubi and Laissue, TIPS, 16, 110-115, 1995).
Symptoms associated with metastatic carcinoid tumors (flushing and diarrhea) and vasoactive intestinal peptide (VIP) secreting adenomas (watery diarrhea) are treated with somatostatin analogs. Somatostatin has been also approved for the treatment of severe gastrointestinal hemorrhages. Somatostatin may also be useful in the palliative treatment of other hormone-secreting tumors (e.g., pancreatic islet-cell tumors and acromegaly) and hormone dependent tumors (e.g., chondrosarcoma and osteosarcoma) due to its antisecretory activity.
Peptidomimetics
As a result of major advances in organic chemistry and in molecular biology, many bioactive peptides can now be prepared in quantities sufficient for pharmacological and clinical utilities. Thus in the last few years new methods have been established for the treatment and therapy of illnesses in which peptides have been implicated. However, the use of peptides as drugs is limited by the following factors: a) their low metabolic stability towards proteolysis in the gastrointestinal tract and in serum; b) their poor absorption after oral ingestion, in particular due to their relatively high molecular mass or the lack of specific transport systems or both; c) their rapid excretion through the liver and kidneys; and d) their undesired side effects in non-target organ systems, since peptide receptors can be widely distributed in an organism.
Moreover, with few exceptions, native peptides of small to medium size (less than 30 amino acids) exist unordered in dilute aqueous solution in a multitude of conformations in dynamic equilibrium which may lead to lack of receptor selectivity, metabolic susceptibilities and hamper attempts to determine the biologically active conformation. If a peptide has the biologically active conformation per se, i.e., receptor-bound conformation, then an increased affinity toward the receptor is expected, since the decrease in entropy on binding is less than that on the binding of a flexible peptide. It is therefore important to strive for and develop ordered, uniform and biologically active peptides.
In recent years, intensive efforts have been made to develop peptidomimetics or peptide analogs that display more favorable pharmacological properties than their prototype native peptides. The native peptide itself, the pharmacological properties of which have been optimized, generally serves as a lead for the development of these peptidomimetics. However, a major problem in the development of such agents lies in determining the active region of a biologically active peptide. For instance, frequently only a small number of amino acids (usually four to eight) are responsible for the recognition of a peptide ligand by a receptor. Once this biologically active site is determined a lead structure for development of peptidomimetic can be optimized, for example by structure-activity relationship studies.
As used herein, a "peptidomimetic" is a compound that, as a ligand of a receptor, can imitate (agonist) or block (antagonist) the biological effect of a peptide at the receptor level. The following factors should be considered to achieve the best possible agonist peptidomimetic a) metabolic stability, b) good bioavailability, c) high receptor affinity and receptor selectivity, and d) minimal side effects.
A generally applicable and successful method recently has been the development of conformationally restricted peptidomimetics that imitate the receptor-bound conformation of the endogenous peptide ligands as closely as possible (Rizo and Gierasch, Ann. Rev. Biochem., 61:387, 1992). Investigations of these types of analogs show them to have increased resistance toward proteases, that is, an increase in metabolic stability, as well as increased selectivity and thereby fewer side effects (Veber and Friedinger, Trends Neurosci., p. 392, 1985).
Once these peptidomimetic compounds with rigid conformations are produced, the most active structures are selected by studying the structure-activity relationships. Such conformational constraints can involve local modifications of structure or global conformational restraints (for review see Giannis and Kolter, Angew. Chem. Int. Ed. Enql. 32:1244, 1993).
Conformationally constrained Peptide analogs
Bridging between two neighboring amino acids in a peptide leads to a local conformational modification, the flexibility of which is limited in comparison with that of regular dipeptides. Some possibilities for forming such bridges include incorporation of lactams and piperazinones. .gamma.-Lactams and .delta.-lactams have been designed to some extent as "turn mimetics"; in several cases the incorporation of such structures into peptides leads to biologically active compounds.
Global restrictions in the conformation of a peptide are possible by limiting the flexibility of the peptide strand through cyclization (Hruby et al., Biochem. J., 268:249, 1990). Not only does cyclization of bioactive peptides improve their metabolic stability and receptor selectivity, cyclization also imposes constraints that enhance conformational homogeneity and facilitates conformational analysis. The common modes of cyclization are the same found in naturally occurring cyclic peptides. These include side chain to side chain cyclization or side chain to end-group cyclization. For this purpose, amino acid side chains that are not involved in receptor recognition are connected together or to the peptide backbone. Another common cyclization is the end-to-end cyclization.
The main limitations to these classical modes of cyclization are that they require substitution of amino acid side chains in order to achieve cyclization.
Another conceptual approach to the conformational constraint of peptides was introduced by Gilon, et al., (Biopolymers 31:745, 1991) who proposed backbone to backbone cyclization of peptides. The theoretical advantages of this strategy include the ability to effect cyclization via the carbons or nitrogens of the peptide backbone without interfering with side chains that may be crucial for interaction with the specific receptor of a given peptide. While the concept was envisaged as being applicable to any linear peptide of interest, in point of fact the limiting factor in the proposed scheme was the availability of suitable building units that must be used to replace the amino acids that are to be linked via bridging groups. The actual reduction to practice of this concept of backbone cyclization was prevented by the inability to devise any practical method of preparing building units of amino acids other than glycine (Gilon et al., J. Org. Chem, 587:5687, 1992). When syntheses of analogs of other amino acids were attempted the method used was unsuccessful or of such low yield as to preclude any general applicability.
In Gilon, EPO Application No. 564,739 A2; and J. Org. Chem., 57:5687, 1992, two basic approaches to the synthesis of building units are described. The first starts with the reaction of a diamine with a bromo acid. Selective protection of the .omega. amine and further elaboration of protecting groups provides a building unit, suitable for Boc chemistry peptide synthesis. The second approach starts with selective protection of a diamine and reaction of the product with chloroacetic acid to provide the protected glycine derivative, suitable for Fmoc peptide synthesis.
Both examples deal with the reaction of a molecule of the general type X-CH(R)-CO-OR' (wherein X represents a leaving group which, in the examples given, is either Br or Cl) with an amine which replaces the X. The amine bears an alkylidene chain which is terminated by another functional group, amine in the examples described, which may or may not be blocked by a protecting group.
In all cases the a nitrogen of the end product originates in the molecule which becomes the bridging chain for subsequent cyclization. This approach was chosen in order to take advantage of the higher susceptibility to nucleophilic displacement of a leaving group next to a carboxylic group.
In a molecule where R is different than hydrogen there is a high tendency to eliminate HX under basic conditions. This side reaction reduces the yield of Gilon's disclosed method to the point where it is impractical for production of building units based on amino acids other than glycine. The diamine nitrogen is primary while the product contains a secondary nitrogen, which is a better nucleophile. So while the desired reaction is generally sluggish, and requires the addition of catalysts, the product is commonly contaminated with double alkylation products. There is no mention of building units with end group chemistries other than nitrogen, so the only cyclization schemes possible are backbone to side chain and backbone to C terminus.
Libraries of backbone cyclized peptide analogs
As mentioned above linear peptides suffer from several serious drawbacks as potential drugs, inasmuch as they are notoriously unstable in vivo, often lack high affinity of binding to their receptor, frequently lack selectively to one kind of receptor, and generally have poor oral bioavailability. In efforts to overcome such problems, it is also possible to utilize the methodologies developed in connection with synthetic peptide libraries to generate collections of cyclic peptides, novel biopolymers and even novel branched oligomeric compounds (reviewed by Zuckermann, Current Opinion in Structural Biology 3, 580-584, 1993).
The generation of libraries of cyclic peptides requires that in addition to any previously stated considerations, the cyclization reaction be performed in a high yield and with a minimum of additional manipulations. Unfortunately, classical cyclization reactions are highly sequence dependent in terms of the expected yields, making the uniform cyclization of a peptide mixture unreliable.
Recent advances in the cyclization of peptides directly on the solid support have improved the synthetic procedure, and even allowed the automation of cyclization reactions based on known cyclization schemes. In the past, cyclizations were typically performed in solution under conditions of high dilution. Polymer-supported cyclizations can both avoid potential side reactions such as oligomerization and facilitate product purification. For example, on-resin cyclization methods have recently been used to prepare cyclopeptides with bridges formed of thioethers, disulfides, or lactams between two side chains, lactam between the amino terminus and a side chain, and lactams between the amino and carboxy termini (reviewed by Zuckermann, Current Opinion in Structural Biology 3, ibid).
The use of resin-bound cyclic peptides and free cyclic peptides in combinatorial libraries is disclosed in WO 92/00091. These cyclic peptides do not contain any conformationally constraining element, however, and in cases where cyclization is achieved, these peptides may still adopt a number of conformations and suffer many of the same shortcomings as linear peptides.
Cyclic semi-random peptide libraries, disclosed in WO 95/01800, are exclusively cyclic penta- and hexa-peptide libraries containing one or more randomized amino acids and a conformationally constraining element in the form of an amino acid residue such as proline which fixes the beta turn angles of the adjacent amino acid residues. The advantages of such conformationally constraining elements is stressed by the inventors of this approach. However, inclusion of such elements via incorporation of a particular amino acid residue into the peptide sequence may have detrimental effects on those residues required for receptor recognition or other biological activity. Furthermore, in that application, the cyclization reaction is merely another coupling reaction in which the terminal amino group of the linear peptide is coupled to the terminal carboxy group of the peptide.